Washington Union Station

Amtrak’s announcement that it needs $7 billion to improve Union Station, in a way that is tangential to train or passenger capacity, has gotten some deserved flak already on other blogs. What I want to discuss instead is a pair of issues relating to capacity: passenger circulation, and track capacity. Especially on the latter, Union Station does have some problems, not at current traffic, but enough that future traffic increases may require difficult at-grade merges. The core of the problem is that the terminal tracks are located to the west of the through-tracks, with an at-grade junction, rather than between them.

Fortunately, the passenger circulation capacity issue is easier. Although Amtrak claims 100,000 passengers use the station every day, in reality the number is beefed up with Metro riders, similarly to Penn Station’s 600,000 daily passengers statistic, of which nearly half is subway ridership. Total ridership on MARC and VRE is 53,000 per weekday, and Amtrak has a total of 13,000 boardings and alightings per day there (not per weekday, but intercity traffic does not have the weekday peak of commuter traffic). This is 66,000 boardings and alightings, assuming every MARC and VRE trip begins or ends at Union Station. In contrast, on just two tracks with ordinary subway platforms, Metro has 34,000 boardings at the station; page 13 of Amtrak’s announcement shows the relative scale of Metro and mainline infrastructure. The mainline half of the station’s ridership is passengers who are likelier to be carrying luggage or not be local, but the main difference between it and the Metro half is that the Metro half is using Metro turf and the mainline half is using the station above which Amtrak’s headquarters is located.

If there is a problem, it comes from Amtrak’s practice of corralling riders at waiting points, instead of letting them filter onto the platforms or the stations whenever they like, as is done every day on trains in France and Germany, or on the less busy stations of the Northeast Corridor. Stephen Smith tells me that unlike in New York or Boston, where the waiting areas are at least adjacent to the platform and the problem is one of having just one access point (or just one official access point in New York), in Washington there is another antechamber between the passengers and the train. An extra 100 meters of walking adds about a minute of travel time in a congested space, and perhaps 45 seconds in a clear one; Amtrak’s current practice adds multiple minutes to door-to-door travel time, and also forces pedestrian congestion once it clears passengers to access the platform.

Adding access points is also a good thing, but that does not cost $7 billion, and does not require redoing the entire main concourse. But possibly the most important thing to do in the near term is making all platforms high, also nowhere near a $7 billion project; the diagrams on Amtrak’s announcement suggest all terminal tracks and most through-tracks will be high-platform, but one through-platform will remain low.

Now, track capacity is where things get more interesting, because potentially there is a problem, coming from terminal layout. A not very clear, but public, diagram can be found here: look for Washington Union Terminal, and within it, Interlockings C (the outer station throat and a nearby yard), K (the inner throat and the actual tracks), and A (the connection from the through-tracks to First Street Tunnel). Note that terminating tracks 7-20 are to the west of through-tracks 22-29, and the junction is at grade, which represents a problem for easy cookie-cutter planning.

The operationally simplest but most expensive to deal with this is to build a grade separation. If it’s anything like Harold, expect a $300 million price tag. At present and expected levels of traffic, this is overkill.

I claim that if MARC and VRE trains continue to terminate at Union Station, no special work is needed: Brunswick and Camden Line traffic can be segregated on tracks 7-9 (and the turnaround capacity, easily about 12 tph for 3 tracks, is more than those lines will need between them), VRE traffic can be segregated on tracks 24-25, and Penn Line traffic can use the same tracks as the terminating intercity trains.

The only at-grade conflict would be between northbound trains originating at Washington, and southbound ones continuing through to Virginia, and even high possible traffic levels (say, 12 tph terminating including the Penn Line sprawled across 11 tracks of which 3 already have long platforms and arguably 3 more can be lengthened, 2 tph through across 4 tracks) can be scheduled in a similar manner to all-terminating stations, treating the through-trains as terminating trains that have to use specific tracks and have no limit on dwell time.

Specifically, because Penn Line (or local HSR) trains would leave immediately after express HSR trains to reduce the number of required overtakes, at worst we’d have trains originating at :00 and :02, repeating every 10 minutes, and then there’s an 8-minute window within which to schedule southbound through-trains.

So instead let us assume commuter trains run through, in which case we may as well assume they have good reliability so that they can be scheduled with 2-minute headways. Current peak traffic is 3 tph Brunswick, 2 tph Camden, 3 tph Penn, and lower combined traffic on the Virginia side. Assume that peak traffic will grow to 3 tph Brunswick and Camden and 6 tph combined Penn and through-HSR; in fact the most potential for growth is off-peak, and because multiple platforms are very long, long trains may be used if there are capacity problems.

We now have 6 tph terminating HSR, 6 tph through-traffic on the Penn Line (including HSR), and 6 tph through-commuter traffic on the Camden and Brunswick Lines; Camden and Brunswick are physically to the west of the Northeast Corridor, and so in addition to conflicts between terminating and through trains, we have conflicts between through-Camden/Brunswick and southbound through-Penn/HSR.

In this situation, we can have southbound terminating HSR and through-Penn/HSR trains clearing the throat at :00 and :02 again. Northbound terminating HSR trains have to depart 2 minutes after the arrival of southbound through-Penn/HSR trains, e.g. :04, and then northbound through-Camden/Brunswick trains can depart between :06 and :08; northbound through-Penn/HSR trains are always to the east of everything else and so do not conflict with anything.

Because southbound through-Camden/Brunswick trains conflict with terminating trains, they can be scheduled at the same time as northbound through-trains of some kind, which constrains the symmetry axis we choose but is otherwise workable. For example, if Camden/Brunswick trains both depart and arrive at :07 then with the terminating trains arriving :00 and departing :04, we have a symmetry axis ending in a 2 or a 7 (and through-Penn/HSR trains would arrive and depart at :02). But then the terminating trains also arrive just before the through-Penn/HSR trains and depart just after, implying they are slower or else there would be an overtake just north of the station. We can instead switch the trains – and then terminating trains arrive and depart :02, and through-Penn/HSR arrive southbound :00 and depart northbound :04. Note that there is no conflict between northbound terminating trains and southbound through-trains.

So it is possible to do this without extra infrastructure beside longer and level-boarding platforms, which are cheap. Let us finish by seeing what extra trains can be scheduled into the above 18 tph schedule. Scheduling 6 tph of terminating trains is easy: trains arriving :04 and departing :00, the opposite of the terminating HSR trains discussed above, will be adequately separated. The problem then is just the need to overtake the :02 through-trains along the tracks; however, at such a level of demand, 18 tph combined HSR and commuter on the Northeast Corridor, full four-tracking there would be necessary anyway.

But no extra through-traffic can be realistically scheduled into the same timetable, because the southbound :04 trains would conflict with the northbound :04 terminating trains. Changing the schedule so that it’s the terminating trains that arrive and depart at the same time is, however, possible: since we’re four-tracking the entire Baltimore-Washington line at this stage, we can have terminating trains arrive and depart :02, Camden/Brunswick trains do the same :07, and through-Penn/HSR trains arrive and depart :00 and :04. That said, this means it’s impossible to schedule more than 6 terminating tph into Union Station; I believe it’ll be easier to fill all those extra intercity trains into Washington than fill 18 tph going from Washington toward Virginia, both intercity and commuter.

Of course, the traffic levels discussed here are all very high, especially for HSR. An HSR system that fills even 6 tph is one that can pay for future capacity increases out of operating profits. The importance should be getting a starter system with reasonable capacity for the next few years and then build capacity projects as required, with immediate construction done only on the most critical segments or those that would be hard to reconstruct with more future traffic.

So we’re back to the question of what needs to be done with Union Station, and the answer is hardly anything. It’s not even Moynihan Station, which is also sold as a bigger transportation benefit than it is, but is at least billed as a grand station to be named after a politician more than anything (and is only about $1.5 billion). It’s even worse than Gateway and the Market East station, which would have positive transportation value, and are just very cost-ineffective. It’s not solving any problem for the foreseeable future; it’s just using big numbers about current traffic and growth to scare people into thinking more capacity is needed, and mostly it’s using small increases in track capacity to justify throwing billions of dollars on beautifying Amtrak’s headquarters.


  1. Ryan

    Washington Union Station needs at least one low platform to facilitate the Capitol Limited, a line on which every single station at present is low platform. Without a low platform at Washington DC, Amtrak would be forced to use high level boarding stock on that line.

    No word yet on why installing high levels in Chicago Union Station and phasing them in on the rest of the line isn’t an option.

    Going by what it’s costing RI to high level Kingston and add a third track ($25 million), we could have that entire line high leveled and passing-tracked for $375 million, which is pocket change compared to some of Amtrak’s price tags…

    • Beta Magellan

      I’m guessing this is mainly a freight issue—leaving CA’s hanging-off-the-side-of-the-boxcar law aside, I’ve read that high platforms can get in the way of a freight train’s dynamic envelope: even if the cars themselves are narrower than the space between platforms, they may still sway back-and-forth enough to cause damage. Since (based on what I’ve heard about the Capitol Limited’s on-time performance) the tracks it uses are freight-primary and there are few passengers boarding at most stations anyway, I doubt it’s cost effective for Amtrak to raise platforms at most stations (especially if we’re including tail and/or passing tracks at stations), though given the small number of passengers boarding and disembarking at most intermediate stations you could probably get away with having an entrance or two accommodating low-level platforms.

      It might also be partly an issue of Metra traffic—all the diesel lines are low-platform and I’ve never seen any proposal for converting all of them to higher platforms for level boarding. (Although I don’t know if Amtrak and Metra actually share any platforms—I’ve only used the north terminus of Union Station and to me it looks like Amtrak has its own platform for the Hiawatha and Empire Builder while Metra uses the rest for the Milwaukee services/NCS with no mingling—but this issue has been gnawing at me for a while.) Some of the busiest Metra lines are also some of Chicago’s busiest freight lines, and tail tracks or extra tracks might be cost-effective. Given the number and volume of Metra stations, folding platforms like on SPRINTER are probably not a cost-effective option. Having mixed high-and-flow floor boarding, like Clem Tillier suggests for Caltrain, probably won’t work because all doors are necessary for disembarking passengers at Union Station (and boarding at some of the busier suburban stations). I wonder whether 500-700 mm might be short enough to accommodate freight (I doubt it) or whether adopting a wider loading gauge might be a way around the dynamic envelope issue (even if this would work, I don’t know if there’s any easy way to accommodate different loading gauges while phasing in—this might require a helicopter drop to fund).

      • EngineerScotty

        Depending on how much space is at the station, that problem (freights passing through striking the platform) is easily fixable with gauntlets or sidings.

      • Richard Mlynarik

        The entire nonsense — and it complete, total nonsense — with freight and high level platforms and gauntlet tracks has nothing to do with 99.9999% of freight, but with a nearly entirely fictional requirement to preserve routes for high/wide out-of-gauge freight movement.

        That means power station primary transformers, aircraft fuselages, and hydrocarbon cracking towers.

        Even thinking about movements of such loads may make sense in Chicago. But such loads are never, under any circumstance, going to move along the NEC or on Caltrain. (If for no other reason than that there are superior water and highway shipping options for every conceivable destination.)

        It is of course trivial to build platforms outside the dynamic loading gauge of every freight car that will ever pass passenger stations that share tracks with freight trains. But because there is absolutely no analysis, especially no cost-benefit analysis, involved by anybody at any level or any stage of any US public transportation project, we end up with purely fictional requirements being cast in stone, by actors (freight RRs, shippers, even sabotaging fellow public agencies such as terminally declining shipping ports) who find they can demand whatever they like, need provide no justification for anything, and can shift any and all costs onto a “public transportation” budget.

        As a final note, there is absolutely no conflict of ~550mm platforms even with out-of-guage loads on well flatcars or on Schnabel transporters. Not relevant to the NEC (where they made the wrong platform height decision 100 years ago, for good reasons at the time), but highly relevant to the western three quarters of the US (where I claim that circa 550mm level boarding into either low-floor single level or the lower floor of double decker trains is the truly optimal solution for all lines, including HSR.)

        So even if there were ever the remotest possibility of high/wide freight movements (which on 90% of the routes there isn’t), there is no conflict at all with level boarding passenger platforms. None except antique regulations, antique standards, and an outright refusal by all parties to do any thinking at all. It’s quite literally insane.

        America’s Finest Transportation Planning Professionals: delivering the worst results at the highest cost!

        • Adirondacker12800

          Freight doesn’t go through Union Station, it bypasses it. The PRR was moderately good at keeping the slow variable traffic – passenger trains – out of the way of the money making stuff – freight.

        • Beta Magellan

          Many thanks, Richard—this was exactly the sort of response I was looking for.

          • Adirondacker12800

            550 mm consigns anyone west of Pittsburgh or Bufflalo to changing trains in Pittsburgh or Buffalo if they want to go east of PIttsburgh or Buffalo. Going to kill ridership on those Toledo to Rochester trips. Or the Cleveland to Philadelphia trips.

          • Zmapper

            I made a map a few months ago of where the existing high platforms are. With the exception of suburban trains in Chicago and Denver (2016), no high platform stations currently exist west of I-81. At the same time, I went through the NEC stations to see if they could accommodate 21″ rolling stock with slight modification. With the exception of New York Penn, every station has enough space to accommodate 48″ NEC and 21″ Long Distance trains.

            As already stated, 48″ platforms tend to interfere with freight movements, and therefore should be limited to HSR tracks or trains that interline with the NEC. It is important to preserve oversize access to the rail lines, because every transformer, windmill, fuselage, etc transported by rail is one less two-lane-wide, hundred-feet-long truck on the road. Of course, there are some lines where accommodating oversize trains is cost-prohibitive (Caltrain, NEC), but along the rest of the rail network said trains already run.

    • jim

      The entire VRE fleet is low-level entry gallery cars. That’s why there are three low-level platform tracks on the east side in this plan. MARC also operates some gallery car trains on the Brunswick line, but I’m not sure for how much longer.

    • Alon Levy

      Ah. For some reason I’d misremembered that the Capitol Limited was a Viewliner train rather than a Superliner train. Anyway, that should probably stay on the terminating tracks because it’s using the Brunswick Line, i.e. the westernmost trackage. Probably the currently unpowered westernmost tracks, 7-9, could be used. (Jim, do you know by any chance which tracks those trains usually use?)

      • jim

        No. The Capitol Limited both arrives and departs during off-peak hours, so can cross as many tracks as it pleases without interfering with other trains. All the other trains supporting checked baggage run on the through tracks, so it might be most efficient (from the point of view of station operations) for the Capitol Limited to use them too.

        • Alon Levy

          Coming to think about it, serious capacity problems at the station are so far in the future that by then they could easily have an HSR train from Washington to Chicago, via Philadelphia. Not the most competitive – it would take 5:30-6:00 – but it would be more competitive than today’s Capitol Limited, and it would also carry Philadelphia-Chicago traffic. Slightly earlier it would still be possible to have HSR from Washington to Pittsburgh with through-service to Chicago with a coupled diesel engine. It’s nothing that we have to think about in 2012, but by the time 6 tph HSR on the NEC are a reality, it’ll be feasible.

          • political_incorrectness

            I was thinking have nodes going to Harrisburg for HSR. Would that include upgrading the entire corridor to 220 mph or using current lines? You could squeeze it down to be competitive when storms effect air traffic.

    • Richard Mlynarik

      Washington Union Station needs at least one low platform to facilitate the Capitol Limited, …

      That’s a lot of “facilitation” (seems like tens of millions of dollars per platform track at Amtrak costs) for one train per direction per day. Or even for 10 times as many Capitol Limiteds.

      Everybody loads up their Christmas list of demands with all the non-negotiable requirements they can dream of, nobody does any cost-benefit analysis, and Somebody Else pays.

      • Adirondacker12800

        Or they could run the Capitol Limited with equipment that can go to low level and high level platforms, like they do with the trains that continue onto the NEC. The foamers in the Midwest a cheap thrill by running retired NEC Amfleets past them…

        • Steve S.

          No way, Superliners or bust. I rode the Cardinal once which has single-level cars and I’ll never make that mistake again.

  2. Max Wyss

    Just FWIW: The Stuttgart 21 project is officially budgeted at 4.3 bn Euro (about $5 bn)…

    • Alon Levy

      Yeah, that’s true. But Stuttgart 21, for all its many, many faults, has transportation benefits (it turns a terminal station to a through-station), requiring extensive reconstruction of tracks underground. Washington 21 has some minor track-area effects, but most of it is the station hall.

      • Richard Mlynarik

        I think you’re missing the point that the Stuttgart21 project cost is not just the new through station, but all the extensive connecting new tracks, with large amounts of tunnelling and high speed junctions and the like.

        http://de.wikipedia.org/wiki/Kosten_und_Finanzierung_von_Stuttgart_21 (Pretty good German language summary of costs and history of cost increases. Wikipedia is quite wonderful for anything I ever care about.)

        Overview on the project’s web site.

        I’m not joking when I say that US costs are three to ten times what they are in “cheap” countries like Germany or Switzerland or Norway. And it’s no joke when the quality of the product is a tenth as good, of worse. (Generally the outcome has negative value.) It’s scandalous, criminal, and heart-breaking.

  3. jim

    This plan is not so much a rail plan as a real estate plan. Akridge has bought air rights (from an arm of USDOT, not Amtrak) over the tracks up to K St and plans to build an entire neighbourhood there: office buildings, a hotel and several residential buildings. That’s one of the driving forces here. To build over the tracks means there needs to be a deck built. Amtrak wishes to widen its platforms, which means realigning tracks (on the west side, there are currently 14 tracks and ten platforms, which will be realigned to 12 tracks and six platforms; on the east side there are 9 tracks and the remains of a tenth which was torn up and five platforms, which will be realigned to eight tracks and four island platforms — three high, one low — plus a low platform the other side of track 29). The deck can’t be built until the tracks are realigned, since supports can’t be placed where there are tracks now or where tracks (or escalators etc.) are intended to be in the future. Both Amtrak and Akridge need the existing garage to be demolished: it’s where Akridge wants to build and its supports are where Amtrak wants to place tracks. So the existing parking requirements and the new parking requirements that will be generated by Akridge’s tenants will need to be satisfied elsewhere; the only available space is below the tracks. If a cavern is going to be excavated to hold the parking spaces, it makes sense to take the opportunity to excavate a bit more to provide additional platform accesses on either side of the parking cavern. MARC, particularly, needs additional platform accesses: the vast majority of its riders terminate/originate at Union Station. If that’s done, then a new concourse needs to be built to connect these additional accesses to the existing Metrorail mezzanine (that’s the Western Concourse in the plan).

    So most of the elements of the plan are either direct consequences of Akridge wanting to build and Amtrak wanting wider platforms or taking obvious opportunities engendered by that necessary work.

    The two elements stemming from Amtrak’s desire to beautify its surroundings — the Central Corridor and the train shed with the undulating roof — are separable and the most likely to be dropped if there’s not enough funding.

    • Richard Mlynarik

      The deck can’t be built until the tracks are realigned, since supports can’t be placed where there are tracks now or where tracks (or escalators etc.) are intended to be in the future.

      That’s not a problem in San Francisco.

      Here they’re spending a total of about $6 billion (the number increases by the day) to build a “train station”, from scratch, in a brownfield, completely cleared, completely under their control site, in which the all of the building structural supports obstruct passenger circulation, all of the building structural support preclude placement of escalators where they are needed, and in which the building structural supports, by their choice. actively prevent high capacity rail service.

      You can’t make this stuff up. But it’s true. And under construction, with your tax dollars. The only solution is, seriously, demolition. It’s that bad. I’m not exaggerating, even if I am intemperately upset about it.

      America’s Finest Transportation Planning Professionals, doing the only thing they know how to do. Screw you over, and suck down billions, while producing the worst results on the planet.

        • Richard Mlynarik

          If I knew what to do I wouldn’t be making blog comments.

          It’s quite tragic, as well as being inconceivably stupid. There are no excuses: no constraints, no political machinations, no secret agendas: just limitless quantities of ignorance, hubris, contempt, stupidity, technical incompetence and deception.

          I just don’t know.

          You could ask the TJPA Executive Director, Principal Engineer and Senior Project Manager what the hell they think they are doing, if you like. (Much of the “decision” making happened elsewhere, of course, most particularly with John Eddy and Tony Bruzzone, now of ARUP) I have asked, and got no answers that were even remotely coherent.

          I simply don’t understand how shit this bad happens. I don’t understand.

          • Zmapper

            Well, are there any representatives likely to side with us on reforming World War II-era rules? My soon-to-be representative Jared Polis (CO-2) seems to be perhaps willing to look into the issue. He has a rail line proposed through his district that is over budget, and the agency constructing it seems to be fairly forward thinking by North American standards.

    • Alon Levy

      The problem is that none of this is worth $7 billion. If Akridge wants to build air rights, he’s welcome to, but doing such an expensive station rebuilding just to facilitate his project is graft that makes Atlantic Yards look like small potatoes. Even the parking spaces they want to add annoy me – they’re planning to massive increase transit access to the station, and it’s already the busiest Metro station, but they still need such a large parking garage? If Washington wants to stop being worse than New York, it needs to stop with the parking bonanza, and realize that if there’s a legitimate need for parking, other people will build garages close to the station, and there’s no need to include that in public projects.

      Most terminating tracks are already Spanish, which means effective platform width is doubled. But if for some reason it’s not enough to just have open-access platforms and also let people board and alight from both sides of the train, they can pave over tracks to create super-wide platforms, and then modify the turnouts to allow trains to access the remaining tracks and also lengthen the platforms when necessary. It’s much easier to do that at Union Station than at Penn Station, because existing traffic is far lower than the turnback capacity of the station, so closing off 2 or 3 tracks at a time is no disruption to service. So with this they can use existing platform locations for a deck. In the interim they directly add escalators and elevators from H Street to some of the platforms; of course if they pave over tracks to widen the platforms they’ll have enough space for two escalators side by side flanking a wide staircase. It doesn’t have to be part of a grander concourse – a simple taxi stand with a shelter, a departure board, and TVMs should be enough.

      For far-future service increases, beyond 18 or 24 through- and terminating tph, they can reserve space for something extra – possibly a grade-separation in the throat eliminating the conflicts, or even a new tunnel from the terminating tracks toward Virginia, since by then they’d need an alternative to Long Bridge. But this is decades in the future.

      • Richard Mlynarik

        OK, drifting off-topic, but I’d not seen bi-sided platforms referred to as the “Spanish Solution” before. And I know this will be Relevant to Alon’s Interests.

        At Madrid Puerta de Atocha they’re implementing another “Spanish Solution” on the long distance high speed terminal tracks: see the second half illustration (“Situación futura”) of page 2 of this document.

        450m platforms with scissors crossovers at the half way point. Arriving 200m long trains discharge at the outer ends of the platforms, and all passenger flow on that half is up from the platforms to a separate departure concourse above the platforms. Trains then proceed to the inner half of the platform (perhaps crossing over to the adjacent platform track), and load passengers. (“En-train” as they love to say at Amtrak. Argh.)

        So passenger flow is uni-directional, with, in effect, separate arrival and departure stations and passenger facilities. Of course much of this is due to the Spanish obsession with fare gates (and baggage scanners!) and the resulting queues and backups and congestion, but some of it is simply down to the huge passenger and train volume they expect, with some reason, to be handling within the next couple decades. The whole thing is designed for heavy throughput, despite the artificial barriers they’ve erected.

      • jim

        It’s not clear who’s paying for what. The Lydia Depillis piece in the City Paper a month or two ago on Union Station quoted Akridge people as though they were going to build the deck. It’s my assumption that USRC and Akridge will pay for building the parking: it’s their requirement. The Master Plan goes out of its way to point out that few Amtrak passengers and even fewer MARC and VRE passengers park at Union Station. Most of the existing parking is used by people visiting the stores in the mall that the original Burnham station has been turned into.

        The stub tracks aren’t really Spanish. Typically it’s a high platform one side and a low platform the other. Passengers get to use one side or the other, but not both.

        As Richard intimates above, once the support structures for the deck are in place, they will prevent any changes to the track and platform structure. It’s best to optimize that structure before the deck goes in.

        Even if the stub tracks could be left alone, the platforms for the through tracks really do need to be completely rebuilt and the passenger access rethought. Right now, track 21 has been torn up. Track 22 is inaccessible. Tracks 27 and 28 aren’t used. Tracks 29 and 30 are inaccessible and track 30 isn’t even powered. Tracks 23-26, the ones that are actually used, are low and narrow with only the one reversible escalator between each platform and a constricted extension of the shed behind the Burnham station that Amtrak uses as its station. I have seen all four tracks occupied simultaneously with Amtrak trains, a consequence of the long Washington dwells. So the currently usable through tracks are actually at capacity.

        • Matthew

          The insane parking requirement is imposed by the government. D.C. and Amtrak have it within their power to avoid forcing Akridge to build a giant parking garage, if they so choose to regain sanity.

        • Alon Levy

          Well, raising the low platforms is important no matter what – the longest platforms today are low, and it’s important to make sure platforms can be lengthened. On top of that, since NY-DC HSR is not that expensive, it’s best to raise and lengthen platforms early, to make sure the new service can use all terminating tracks.

          The cost of this is nonzero, but the location is not all that constrained because of the surplus of stub tracks today, so it should be ~2 orders of magnitude cheaper than Amtrak’s proposal, at least. Most comparable US projects I know of involve infill stations built from scratch along active commuter lines, with much more construction than just platforms. But raising both platforms at one Fairmount Line station was $6.5 million.

          I don’t know what present practices are for through-trains. But because HSR works with permanently coupled trainsets, even when power cars are involved, through-service to unelectrified territory means coupling extra diesel locos rather than uncoupling electric engines and replacing them with diesels, so it should be faster than 30 minutes. At any rate, if the cost of redoing the platforms is this high, then they can electrify to Lynchburg and either Norfolk or Newport News and move to the LD engine change point to Richmond (or Lynchburg for the Crescent); it would improve VRE service, too.

          • Richard Mlynarik

            Here’s a random comparative platform-raising cost I stumbled across, as one does (doesn’t one?):
            Arth-Goldau station in Switzerland. Raising and lengthening two island platforms and two other platform faces (map) to 420m and level boarding at 550mm.

            Total project cost CHF6m (which is over $6m USD, given their terrifying exchange rate), construction period.

            Switzerland isn’t a cheap country(!), and this station sees several hundred trains a day, passenger and all-day very heavy freight traffic. (It’s “GD” in this diagram, which shows only one part of the traffic.) The project involves complete renovations of the platforms and all passenger facilities.

            So, looks like the MBTA project you cited is costing at least 3 to 4 times as much as it should, possibly ten times.

            Why can’t we have nice things?

          • jim

            DDOT did a study of Union Station a couple of years ago and proposed raising one of the lower level platforms (the one between tracks 25 and 26) in place. The cost estimate in the study for that effort was $540K.

          • anonymouse

            @Richard, because SBB does this sort of thing all the time, while the MBTA does it quite rarely, so SBB has a very good idea of how much this sort of thing it should cost and how to design and manage the project so as to minimize costs, while the MBTA does not. Also, the scope of work at that MBTA station sounds not too different from what you describe: two high level platforms, full renewal of all passenger facilities, such as they are, and probably construction of new access ramps. So it’s really more like 2-3x more expensive, which is hardly an order of magnitude difference, but it’s still a lot. At least the MBTA is doing it though, and it’s a step in the right direction toward making this line into an S-Bahn type of operation.

          • Max Wyss

            @anonymouse: Yes, the SBB does such projects all the time, and they do have their own departments to do such projects, which means that it is the SBB running the show, and not consultants.

            Raising the platforms to 55 cm is/has been a common task, because of the Swiss accessibiltiy laws, and the introduction of low-floor vehicles (which do not have steps anymore). In the S-Bahn Zürich area, it was also necessary to provide a solution because the rolling stock requires 55 cm platforms.

            A complete rebuild of an existing station with a 300 m island platform and one access with stairs and ramps plus a new station building and remake of the station square costs something between CHF 6 and 10 millions. And that does include repositioning (and essentially replacing) the tracks.

            @Alon: Unless they procure a series of hybrid trainsets (à la Talgo, or the new British sets), it will mean coupling a diesel locomotive to the train for through trains, or run the through-trains as “fast train” with conventional rolling stock, and provide a timed connection with an “express train” on the same platform. Of course, electrifying would give a serious boost to the VRE services as an intended side-effect.

    • keephoustonhouston

      Viewed through another lens, this kind of illustrates the folly of building massive structures over transportation networks that aren’t themselves ridiculously overbuilt.

      DC is facing this now with the parking structure. Likewise, Seattle has long faced it with the Convention Center over I-5. You wanna see how to do that shit properly, check out the fifth runway at ATL – the underpasses for the Perimeter are fully DOUBLE the current lane width, conceivably accomodating a 20-lane cross section.

      Not that it’s going to be 20 lanes tomorrow, but if you’re going to subordinate all future transport decisions to the stuff above, that’s what you gotta do.

  4. political_incorrectness

    What is the standard width on platforms built at HSR stations like St. Pancras UK, Puerta de Atocha etc?

    • Richard Mlynarik

      About 9-10m for an island is the low end of what people aim for. It depends on volume, and the heaviest traffic is, of course, associated with local, not inter-city, traffic.

      Stuttgart21 (projected traffic: 75000 pax/platform/day) will be 10m. (Three access points with along each platform, each with multiple and direct routes to the surface; total of five banks of escalators/stairs and three elevators per platform.)

      The new underground (mostly S-Bahn, with a decent helping of intercity) through underground platforms at Zürich HB (Löwenstrasse) will be 13m (up from 10m of the existing underground through platforms.) Serious volume.

      I’m reading off some low-res scans, but to the best of my estimation the new under-construction (and massive, and massively impressive) Barcelona La Sagrera‎: has 10m wide islands for both AVE (HS) and rodalies (S-Bahn).

      PS I wasted an insane amount of pointless time a while back making this multi-layer PDF out of La Sagrera‎ plan scans. “Time management skill” is my middle name. (Note that Apple’s PDF browser, and perhaps many others, don’t do the PDF layer thing. You’ll probably need Acrobat.)

      • Max Wyss

        From my personal experience in Zürich, the 10 meters in the Bahnhof Museumstrasse is a minimum for S-Bahn use (made worse by the sloooow escalators); that’s why they went to 13 m in the Bahnhof Löwenstrasse. Another reason for the wider platform in the Bahnhof Löwenstrasse is that service vehicles (baggage, restaurant car supplies) have to circulate on the platform as well.

        About the PDF: Apple’s Preview.app is an inadequate viewer; it fails miserably when validated against ISO 32000. Adobe Reader (or, of course, Acrobat Pro) is (still) the standard viewer (to be used for documents “better” than simple plain PDFs).

  5. Zmapper

    On the west side of the main concourse, Amtrak’s plan to align the existing tracks into 10 tracks fronting 5 wide platforms seems like a worthy investment. Perhaps half of the parking garage could be reconstructed to fit the new platform dimensions, followed by the other half. In the mean time, existing auto passengers could access the station by Metro, or if the absolutely insist on driving a temporary parking lot could be set up at RFK Stadium and the streetcar provide shuttle access to Washington Union.

    On the east side, Amtrak proposes two terminating tracks, followed by 8 tracks fronting 4 platforms, at a lower elevation to the other tracks. The two terminating tracks closest to the central concourse should be lowered to the same elevation as the through tracks, and should be connected to the First Street Tunnel at the southern end. 10 through tracks would then exist, providing ample track space for existing and future needs.

  6. simple

    In May of this year the City of Chicago released a Master Plan for its Union Station (in cooperation with Amtrak, the station’s owner, and Metra, the station’s busiest tenant) which proposed a lot of low-cost, seemingly sensible, yet significant capacity-increasing improvements in the “medium term” and pushed the more fanciful billion+ dollar tunnels, etc. out to the “long term.” http://www.unionstationmp.com/ Interestingly, over the past months since its release it has received but a small fraction of the blog commentary that the Washington Union Station plan has in the past two weeks… Maybe more sensible plans just aren’t good fodder for discussion?

    • Beta Magellan

      Well, it’s not nearly as egregious.

      I’d say the plan’s good if we accept current operating practices (parking trains at downtown terminals, low-off-peak frequency diesels 4EVA) as good. If you want to see a revised Union Station as the anchor for a more RER-ish regional rail system, it’s quite disappointing (although Metra through-routing has always been of secondary concern for these plans—the possibility of putting HSR through a new tunnel seems to usually take precedence).

      • keephoustonhouston

        Watching SEPTA systematically dismantle Vuchic’s numbering system has convinced me that any attempts to bring RER-style through running to America, no matter how good on paper, will ultimately end in doom.

        • Adirondacker12800

          SEPTA decided that running empty trains to satisfy a through running fetish wasn’t a good idea.

  7. Andre Lot

    On recent weeks, I’ve been left quite astonished how the transit blogosphere has somehow become repeating again and again arguments that assume two extremely unlikely things to happen in North America:

    – adoption of dumb clockface scheduling practices in US that assume no transit app, no smartphones and some sort of mythological, almost religious worship of the number 60 and its natural divisors.

    – some paradigm shifting that would adopt the “build-to-schedule” practices of the Swiss as if they were some sort of universal panacea for high construction costs in US (only other small countries follow the Swiss practice on this, the whole high-speed networks of France, Italy and Spain were not designed with tentative schedules to the minute in mind).

    • Richard Mlynarik

      You’re going to have to avoid France. They’re getting ““dumber” by the hour.

      Stick to retarded third world countries, like the US. Oh, except the US also has wretched mobile data, and guaranteed inaccurate vehicle tracking, so good luck with that whole transformative smartphone business.

      Now excuse me, I’m off to catch BART here in San Francisco, departing :02 :09 :17 :24 :32 :39 :47 :54.

      • Andre Lot

        My argument stays: in rich countries, transit planners can just ignore the cavemen technophobes and disregard them. Therefore, one could use some optimization routine to find what would be an optimal interval for a given infrastructure. It might be 2234 seconds, it might mt 3600 seconds exactly, it might be 3914 seconds… who knows – the computer.

        (I bet there are some cashiers out there on outdated stores who “proud” themselves on making mental summing of grocery bills instead of scanning their bar codes)

        • Zmapper

          So what do you do for the people who have a dumbphone, like me, or just don’t have a mobile phone? Tell them to hope that a bus might possibly be coming soon?

          • orulz

            Simple: just distribute printed schedules.

            When I lived in Hiroshima I rode the bus a lot. Buses in Hiroshima followed extremely irregular schedules. They actually posted full schedules at each and every stop, and distributed business card-sized schedules for each route that fit conveniently into wallets. The routes that I rode all served a particular university; this allowed the bus company to be very flexible, and tailor the schedules to the exact times that classes began and ended, which definitely did not follow a clockface schedule.

            That’s sort of a special and extreme case, though; but even the JR in the region didn’t follow an all-day clockface schedule. The schedules would sometimes be completely different from hour to hour with several different service classes. They solved this irregularity by posting printed schedules all over the place, in a very easy to understand format (which I have not seen duplicated anywhere in the US.)

            In any event, I agree that rigid adherence to clockface scheduling, though perhaps nice, isn’t really as all-important as folks like Richard M., and Alon L., and Clem T. make it out to be (From reading some blogs, you might get the impression that if you are a planner and you implement clockface schedules your IQ is normal, if not, it falls somewhere in the range of -25,000).

          • Alon Levy

            So, to be fair, my knowledge of Japan comes 100% from trolling Hyperdia for Tokyo-area schedules. There there do exist clockface patterns on many lines – the Kodama has something like a 2 tph takt, the Tokaido Line has 5 midday tph in a generally repeating pattern, etc. The busiest lines have nothing like it, which I’ve always handwaved as a “When you run 15 tph until midnight, nobody gives a crap about the schedule” issue.

          • Andre Lot

            I think there shall be two networks (that might overlap) concepts:

            (i) high-frequency (or low interval), based on spaced dispatching (e.g., keeping the intervals at each stop/station regular instead of forcing the operations to adhere to a scheduled time), which, applied within a networks, dramatically reduces waiting time at the beginning and at transfers – truly ‘turn up, go and forget about schedules’

            (ii) scheduled services, for which people are expected to check some external source for their trips (printed, phoned – think of a voice information in which you type line code, stop code and some automated voice machine spells the next 3 or 4 stops, online on the Internet, VPD etc.)

            Services defined in (i), if really highly frequent (4 min or less) operate better under controlling the interval between vehicles than sticking them to a schedule – they are more short-term responsive. Services defined in (ii) probably need an integrated network scheduling but it doesn’t need to be based on an interval based on some magic number like 60 minutes or its divisors.

            It looks like the “clockface scheduling” is one of those fanboys geeky concepts that poison the real World of complex operations. As I wrote, I feel like almost a cult: there is NO reason, in the age of microprocessors and electronics, to take decision of pouring concrete and massive amounts of money based on some almost religious allegiance to the number 60.

          • Richard Mlynarik

            Depends in the end on whether you think Switzerland or Berlusconiland (or Somalia, for that matter) has a better transportation system.

            Actually, it isn’t a matter of what one thinks: one can look at the numbers. Crazy fetishistic behaviour.

            Or one can posit a smartphone-enabled future along with asteroid mining and fusion powered space flight.

          • simple

            @ orulz

            CTA prints and posts rail schedules that are a modified version of the Japanese standard (hours on the left column with arrivals in minutes past each hour in columns to the right). CTA used to list all arrival times, as in Japan, but several years ago switched to the current format (intervals only during frequent times, see for example: http://www.transitchicago.com/assets/1/orangeline_schedules/Halsted.pdf )
            with the logic being that CTA’s trains don’t follow the schedule to-the-minute during frequent times (unlike in Japan), so why imply to riders that they do.
            Of course there’s always CTA TrainTracker too for the vast majority of transit riders who aren’t slow-adopters of technology. See http://www.transitchicago.com/traintracker/arrivaltimes.aspx?sid=41130
            for the equivalent real time info for the station in the printed schedule link above…

          • simple

            @ Alon

            You really need to visit Japan in person! You’ll be even more impressed than by reviewing schedules online (which is indeed also quite impressive). But I think with some in person operations of amazing high tph/track terminals like the Chuo Line at Tokyo or the Toyoko Line at Shibuya (just the tip of the iceberg) you will also gain a greater appreciation for the operations discipline it takes to live up to those impressive schedules (and it should be obvious then all sorts of reasons why assuming the same or similar can be achieved in the US, while theoretically possible, is probably far-fetched)…

        • Adirondacker12800

          Every 37 minutes? Every hour is easy to remember but transit that runs once an hour does have much demand does it? Every 65 minutes wouldn’t work well at all.

        • keephoustonhouston

          I intentionally own a dumbphone because the times when I’m away from a computer are my respite from the urge to catch up with whatever inane bullshit is being posted on facebook, reddit, or whatever.

          It’s one thing to say we shouldn’t fetishize 30-minute intervals the way the Swiss do, but it’s not too hard to round to the nearest 5-10 minutes. Every 40 minutes, or every 20, or every 15. For awhile in Houston I was using a bus route that ran perfect 18 minute headways. All afternoon, every 18 minutes. Even that you could do in your head.

          As far as doing math in your head goes, all exercise is good – not just the physical kind.

        • Henry

          You do realize that it is possible for mobile communications systems to fail, right? An app is useless if say, another 2003 blackout or another network overload like 9/11 causes the loss of cell service.

          The other thing is that natural divisors of 60 also happen to be lengths of time people are willing to wait for a service. Anything less than 10 minutes between services is the point where you can just show up and wait. If they miss a transfer to a bus or a train, they might be willing to wait 12, 15, 20 minutes, but i highly doubt that anyone really wants to wait 27 minutes for the next bus after missing one.

          Lastly, you are assuming that American transit agencies are competent managers with the resources to devote to running service efficiently, which they are not (the NYC MTA being a case in point)

          • Adirondacker12800

            The schedules for electric trains are pretty useless during a blackout. Schedules for diesel trains, during an extended blackout, too.

    • Alon Levy

      I’m assuming clockface symmetric schedules because it’s easier to plan to them when you have my level of access to complex scheduling software (i.e. none). The point here is not that Amtrak should adopt an integrated timed transfer system, although it should; it’s that you can schedule trains at high capacity with the existing flat junctions and with something like the existing infrastructure, without building a Washington 21-style project.

        • Alon Levy

          With other intercity lines, though Washington isn’t much of a junction. (In the far future, Cleveland is; in the less far future, Albany is.)

          But potentially also with commuter lines – intuitively the frequency of MARC, VRE, and Amtrak feels like it should be similar, and much higher in the off-peak than it currently is. So trains could still more or less pulse, at traffic levels higher than today and lower than the limiting case I described in the post.

  8. Pingback: Low- and Medium-Hanging Fruit | Pedestrian Observations
  9. Pingback: When Through-Running Is Inappropriate | Pedestrian Observations
  10. Pingback: Why Northeast Corridor Privatization is Doomed to Failure | Pedestrian Observations
  11. Pingback: No New Washington Union Station, Please | Pedestrian Observations

Leave a Reply to Alon Levy Cancel reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.